Sourcing Food Influencers Is 10% of the Work. Nobody Prices the Other 90%.
Sourcing is only the start. See the booking, visit, file, rights, and repeat-work system restaurants need to manage food influencers.
Most of the work of managing food influencers happens after you choose who to work with. In Remy's 2026 records, 94.2% of text messages tied to completed influencer visits happened at or after approval.
This is the management problem underneath influencer marketing for restaurants. Finding someone is necessary. Getting that person from approval to a finished visit, complete delivery, usable rights, and a possible second invitation is the work that makes a program repeatable.
“Sourcing is 10% of the work. Management is the 90% nobody prices.”Feast operating thesis, August 2026
The 10/90 phrase is Feast's shorthand for that workload, not a time study. The message count supports the point about follow-up, but it doesn't measure how many hours an owner spent.
Sourcing looks like the whole job because it is the first visible win
The old view was reasonable: without an influencer, nothing else can happen. Search results, databases, DMs, and applications all put the emphasis on finding a name, so an approval can feel like the hard part is over.
Approval actually creates a new set of obligations. Someone must get a date on the calendar, make sure the terms are understood, confirm the visit, prepare the staff, handle a reschedule, record whether the person showed up, collect the post and files, review what arrived, store the rights, and decide whether the relationship is worth continuing.
That difference matters because sourcing and management fail in different ways. Too few suitable applicants is a sourcing problem. Plenty of approvals followed by unbooked visits, no-shows, missing files, or unclear rights is a management problem. Buying more names does not repair the second one.
What changed is that a restaurant can now publish one offer, let interested local influencers apply, and keep the downstream work in one record instead of treating each DM as a new project. Once suitable applicants arrive, booking and follow-up become the work that keeps them from falling through.
Approval starts the expensive part
An approved influencer is not a booked visit, and a booked visit is not finished work. Every stage needs its own status so a busy owner can see the first place the work stopped.
“Someone has to recruit a creator, the creator visits, they produce content, someone has to review it, someone has to give feedback, someone has to revise.”Feast product-team note, April 2026
You need to know who is booked, who has visited, and who still owes you content. Otherwise an approved application can look like a completed collaboration even while the visit and video are still unconfirmed.
The full sequence is the subject of the seven-step loop behind every influencer visit. The important shift here is that the useful unit is not an influencer you found. It is a collaboration that delivered the work in the agreement.
Decide whether to do it yourself, rent access, or own the network
There are three ways to carry this work: do it yourself, rent access to a network, or build a direct network with a management system. The most efficient choice depends on whether this is a one-time campaign or work you expect to repeat.
| Choice | What your team handles | What you retain next month |
|---|---|---|
| Do it yourself | Search, terms, booking, reminders, reschedules, visits, files, and follow-up | The direct relationships, records, files, and licensed uses you organized yourself |
| Rent access to a network | Whatever recruiting, booking, and follow-up the provider does not include | Whatever contacts, history, files, and rights the contract allows |
| Own the network with a management system | Influencer approval, content judgment, and real exceptions | Direct relationships, completed-work history, delivered files, and only the uses granted in writing |
Rent and own describe business models, not a provider's legal power to stop contact. The contract decides contact, export, repeat-booking, file, and license consequences. For a single campaign, rented access may be efficient. For monthly work, an owned network with management software is the more efficient fit when avoiding another blank search and preserving completed-work history matter. DIY has the lowest platform cost and the highest restaurant workload.
Agree on the post and files before anyone books
Management gets easier when the offer tells both sides exactly what completion means. Write the terms before approval turns into a calendar problem.
At minimum, record:
- the restaurant, date, time, guest count, and staff handoff;
- the food credit and any agreed cash compensation;
- the live-post requirement, platform, deadline, and disclosure instruction;
- the finished files, raw footage, aspect ratio, and revision process;
- the commercial-use license for advertising from the restaurant's own account;
- separate account-level permission only if an ad will run from the influencer's account;
- any bonus that becomes due only when the restaurant actually uses the content in paid advertising under the written terms; and
- the person responsible for approving the influencer and the delivered work.
A file is not a commercial-use license, and a license is not copyright ownership. Account-level partnership-ad permission is a separate approval needed only when the ad runs from the influencer's account. An actual-paid-use bonus is payment under the written terms; it does not grant that account permission. The payment distinctions are covered in how restaurants can pay for useful influencer work.
Keep ordinary questions out of the owner's inbox by writing default answers for arrival, guest count, menu exclusions, rescheduling, file upload, and deadlines. Send only real exceptions to the owner: a questionable applicant, a material schedule conflict, a disputed deliverable, or work that needs judgment.
Count the work without pretending messages are hours
The clean report shows where the work accumulated, not a made-up dollar value for every text. Count the records first; add an owner-time estimate only if you write down a real minutes-per-task assumption and keep it labeled as your planning estimate.
The 2026 Remy message cohort makes the downstream load visible. It recorded a median of 25 SMS messages per completed visit, with 94.2% of the messages at or after approval. In that same cohort, 3,839 of the 8,540 messages were timestamped at or after the scheduled visit start, when post-visit questions and file follow-up can begin; some of that traffic was automated.
Those are message-event counts, not a claim that an employee typed every text or spent a fixed number of minutes on it. They also say nothing about whether the content caused a guest visit, sale, or profit.
Each week, check the upcoming visits and overdue content. Keep future bookings and work still within its deadline separate from actual missed commitments.
The source of a miss then becomes obvious. Low application volume calls for a different recruiting approach or offer. Approvals that never book call for faster scheduling. Bookings that become no-shows call for confirmations and a clear exception path. Completed visits with overdue files call for a written deadline and follow-up. Complete files without useful rights call for better terms before the next visit.
Remy is useful when the middle keeps breaking
Remy becomes relevant after you have a clear offer and approval rule but cannot reliably carry every approved influencer through booking, reminders, rescheduling, visit follow-up, and file collection. A local recruitment ad shows nearby influencers your restaurant's written offer; interested people apply, and you approve or decline them. Remy handles the repeated coordination in between while you keep the decisions about who comes and what content is acceptable.
The documented Remy sequence makes that manager's work concrete: booking confirmation, a confirmation request before the visit, an arrival reminder, a post-visit file request, and later follow-ups when delivery is still open. The automation does not turn an approval into a completed visit or a submitted file into approved work. It moves the record forward and leaves exceptions visible.
Remy is the influencer recruiting and management product. The full Feast service is separate; it can run restaurant Meta advertising and connect supported marketing records to sales. A Remy message or completed influencer visit is not proof that an influencer post produced revenue.
Price the manager, not just the introduction
The better buying question is not “How many influencer names do I get?” It is “Who makes sure each approved influencer shows up and delivers the promised content, and what will my team still own next month?”
Compare the software or service fee, recruitment spend, food, and any separately agreed influencer or content-use payments. Ask how much work remains with your team. For a detailed budget, see what Remy costs.
Sourcing deserves attention until suitable applications arrive. After that, close the booking, visit, delivery, rights, and repeat-work loops without hiding the exceptions. That is why the 10/90 thesis changes the purchase: the valuable tool is the one that manages what happens after yes.
Remy is $300 a month per location: $150 for Remy and $150 for creator recruitment ad spend. There is no per-influencer software charge. Hosted food, agreed cash compensation, extra rights, account-level permission, actual-paid-use bonuses, and customer-facing media are separate. Learn more about Remy.